Information Technology Reference
In-Depth Information
those features in cross-cultural communication. The virtual agents, however,
are unsuitable for interactions in which ECAs have to interact with substantial
objects in real world because most of them do not have their body in real world.
On the other hand, interactions which are suitable for robots are vice versa. For
example, we had better use robots to investigate a multiple-party interaction in
which people in different cultures take part because virtual agents cannot join
the multiple-party conversation. Both these kinds of ECAs are used depending
on the situation. Therefore, we develop CEBE to be able to use both types of
ECAs and we will investigate them in both situations.
3.1 Necessary Conditions for CEBE
CEBE is an environment in which a human can interact with another human who
has the appearance and physical limitations of ECAs. CEBE has two aspects; one
is a system to obtain data of cultural behavior in a human interaction through
a body and perception of an ECA. Another is a system to operate a body of an
ECA by using perception of the ECA and obtained data.
In either case, a person in CEBE has to be able to perceive the same level
of information which an ECA can perceive. We need to consider the difference
of information processing between ECAs and humans. For example, a human
can recognize his/her surroundings by using only video in most cases, because
visual information is easily interpretable by humans. ECAs, however, need lots of
sensor data, such as a range sensor, 360-degree camera, microphone, and so on,
to recognize their surroundings at the same level. It is important to provide both
a person and an ECA with the same level of information. It is, however, dicult
to process all of the information which is used in human-human communication.
Therefore, we need to control the information provided to a person depending
on the situation.
We also have to be able to obtain and express cultural interaction behavior
in a human-agent interaction. For example, we behave in various ways, such as
shaking our heads, making hand gestures and changing prosody, when we ne-
gotiate with others. We also pay attention to nonverbal behavior to speculate
others' intentions. We expect that significant information which expresses cul-
tural interaction behavior is gaze directions (head orientation), movements of the
upper body and hands, and prosody of utterances (pitch and power of voices).
The accuracy of measuring and reproducing behavior is dependent on the sit-
uation. We expect that ECAs cannot completely reproduce human behavior in
most cases. We also expect that people can convert their cultural behavior which
ECAs can reproduce by mutual adaptations, in the case that they understand
the appearance and physical limitations of ECAs.
To sum up, necessary conditions for CEBE are as follows:
-
Both a person and an ECA are provided with the same level of information
which is needed to interact with cultural behavior.
-
Cultural interaction behavior in a human-agent interaction can be obtained
and expressed.
